On 8/05/2012 5:19 AM, Elton Carvalho wrote:
Thank you for your reply, Mark,
On Sat, May 5, 2012 at 7:06 AM, Mark Abraham<mark.abra...@anu.edu.au> wrote:
Fundamentally, pdb2gmx is built for fine-grained force fields. I'd have
supposed Martini has some kind of builder program other than pdb2gmx, for
this kind of reason, but I have no idea whether this is true.
It kinda does. Not for polymers just yet. I asked part of the people
involved in this before asking here.
Setting nrexcl = 1 would exclude such angles and dihedrals, I think, but
would also exclude generating the normal ones. One option is to do that and
explicitly define the relevant angles and dihedrals.
nrexcl is already 1, as per MARTINI's defaults and the dihedrals are
built anyway.
Yeah, I was wrong in that speculation. nrexcl pertains only to
non-bonded interactions.
Code comments are in disagreement about column 8. Looking at the code, I
think the above .rtp comment is how the code works. A specific simple
counter-example (showing how .rtp contents create .itp contents that seem to
breach the above description) would be welcome. I'm not sure if there are
circumstances where impropers are generated automatically, but that could
possibly explain such an observation.
I prepared a minimal example and, in doing so, I noticed that the
problem is not removing dihedrals sharing a bond with impropers, but
column 5: all_dihedrals, which is said to generate dihedrals involving
ony "heavy" atoms if set to 0.
When set to 1, pdb2gmx generates the dihedrals CA2 CA1 C9 C91 and CA1
C9 C91 C4 for example (see attached figure, remembering this is a
coarse grained model), which are not generated when all_dihedrals is
zero. Certainly, none of these atoms should be considered "light".
This raises a question about how "heavy" atoms are defined in Gromacs.
I couldn't find it in the PDF manual. Where should I put the
definitions of which atoms are "heavy"?
pdb2gmx looks at whether the atom name starts with 'H' or a digit and
then 'H', but it turns out not to be relevant.
Also, regardless of the value of RemoveDih, dihedrals like CA2 CA1 C9
C91, that shares the central bond with CA2 CA1 C9 CB1, are kept in the
final topology. So apparently RemoveDih doesn't really exclude proper
dihedrals sharing the central bond.
Yeah, that .rtp comment looks totally wrong. After looking at the code I
think:
* Column 5: 1 means keep all generated dihedrals,
* 0 means permit generated dihedrals to have their parameters
* superseded by ones on the same central bond that have
* fewer hydrogen atoms.
That's totally different from what the .itp suggests, and does explain
your observations.
There seems to be no automatic way to treat the issue that you'd like
addressed. Feel free to make a feature request on redmine.gromacs.org,
but it likely won't be addressed until GROMACS 5 (at least).
I prepared a minimal structure and a gmxlib tree which includes the
RTP file for clarification, but it exceeds the attachment limit of the
list, I can send it if its' really necessary
Could anyone clarify the role of these extra columns? I think this
also hints for a review in this section of the manual.
True. An update is in progress now.
That's great news!
In the next days I'll send some questions about the manual to help
guiding that effort! ;)
There have been quite a few manual updates already. For example,
http://jenkins.gromacs.org/job/Manual_Gerrit_4_5/15/ has a link to a
current PDF manual ("save as" on the PDF link is sometimes necessary).
Feedback is welcome, but feedback relative to the updated manual is more
welcome. :-)
Mark
Greetings from a sunny Groningen,
--
Elton Carvalho
Faculty of mathematics and natural sciences
University of Groningen
--
gmx-users mailing list gmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
Please don't post (un)subscribe requests to the list. Use the
www interface or send it to gmx-users-requ...@gromacs.org.
Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
